Solving the Optimal Stopping Problem

This is a presentation I gave at an MLH team building event. The goal was to teach the team something new in 3-5 minutes. I decided to focus on the Optimal Stopping Problem, which is an algorithm for determining when to make a decision when you have incomplete information.

Transcript

1. SOLVING THE OPTIMAL STOPPING PROBLEM
2. How do you make the best decision with incomplete information? Personal Life Apartment Hunting Dating Parking Spots Searching on Seamless Job Hunting MLH Life Member Events Fellowship Admissions Recruiting Hackathon Judging Researching Vendors
3. With time we gain a better & more complete picture. The problem is, you can’t just search indefinitely.
4. Scenario: We’re looking for a house to buy. We plan to see 10 houses. After viewing each house, we have to decide yes or no. We can only look at the next house if we say no to the current one. Our goal is to end up with the best possible house. What is the best strategy? Score ? ? ? ? ? ? ? ? ? ?
5. Score 3 13 ? ? ? ? ? ? ? ? If our sample is too small, we don’t get enough information to compare data points.
6. Score 3 13 17 4 15 19 22 5 10 2 If our sample is too large, we’ve burned through too many good choices.
7. THE ANSWER IS 37 %
8. Score 3 13 17 4 15 19 ? ? ? ? Spend 37% of your time looking (no buying!) & then buy the first thing you see that’s better.
10. How do you make the best decision with incomplete information? Our Process Know the Sample Size Get a baseline Make decisions
11. WE HAVE REACHED OPTIMAL STOPPING MOMENT FOR THIS PRESENTATION.